Polyhedra: Plato, Archimedes, Euler

Transcription

1 Polyhedra: Plato, Archimedes, Euler Robert L. Benedetto Amherst College MathPath at Mount Holyoke College Tuesday, July 15, 2014

2 Regular Polygons Definition A polygon is a planar region R bounded by a finite number of straight line segments that together form a loop. If all the line segments are congruent, and all the angles between the line segments are also congruent, we say R is a regular polygon.

3 Polyhedra Definition A polyhedron is a spatial region S bounded by a finite number of polygons meeting along their edges, and so that the interior is a single connected piece. So these are not polyhedra: But these are:

4 Regular Polyhedra Definition Let S be a polyhedron. If all the faces of S are regular polygons, all the faces of S are congruent to each other, and all the vertices of S are congruent to each other, we say S is a regular polyhedron. Cube (Regular Hexahedron) Regular Octahedron

5 Another Regular Polyhedron Regular Tetrahedron

6 Two More Regular Polyhedra Regular Dodecahedron Regular Icosahedron

7 And They re Dual:

8 Some Ancient History (of Regular Polyhedra) Pythagoras of Samos, c.570 c.495 BCE: Knew about at least three, and possibly all five, of these regular polyhedra. Theaetetus (Athens), BCE: Proved that there are exactly five regular polyhedra. Plato (Athens), c.426 c.347 BCE: Theorizes four of the solids correspond to the four elements, and the fifth (dodecahedron) to the universe/ether. Euclid (Alexandria), 3xx 2xx BCE: Book XIII of The Elements discusses the five regular polyhedra, and gives a proof (presumably from Theaetetus) that they are the only five. For some reason, the five regular polyhedra are often called the Platonic solids.

9 There Can Be Only Five: Setup Given a regular polyhedron S whose faces are regular n-gons, and with k polygons meeting at each vertex: 1. k 3 and n S is completely determined by the numbers k and n. 3. S must be convex: For any two points in S, the whole line segment between the two points is contained in S. 4. Since S is convex, the total of angles meeting at a vertex of S is less than 360 degrees. ( 180(n 2) ). 5. Each angle of a regular n-gon has measure n [Triangle: 60, Square: 90, Pentagon: 108, Hexagon: 120.]

10 There Can Be Only Five: Payoff From Previous Slide: The regular polyhedron S is determined by n 3 (faces are regular n-gons) and k 3 (number at each vertex). But total angles at each vertex must be less than 360. Triangles (n = 3): k = 3 triangles at each vertex: Tetrahedron k = 4 triangles at each vertex: Octahedron k = 5 triangles at each vertex: Icosahedron k 6 triangles at each vertex: Angles total 360. NO! Squares (n = 4): k = 3 squares at each vertex: Cube k 4 squares at each vertex: Angles total 360. NO! Pentagons (n = 5): k = 3 pentagons at each vertex: Dodecahedron k 4 pentagons at each vertex: Angles total > 360. NO! n-gons (n 6): k 3 n-gons at each vertex: Angles total 360. NO!

11 The Five Regular Polyhedra

12 But What About This Solid? Cuboctahedron Known to Plato Definitely not regular (squares and triangles), but: All faces are regular polygons, and All the vertices are congruent to each other.

13 Semiregular Polyhedra Definition Let S be a polyhedron. If all the faces of S are regular polygons, and all the vertices of S are congruent to each other, we say S is a semiregular polyhedron. [Same as definition of regular polyhedron, except the faces don t need to be congruent to each other.] Natural Question: How many (non-regular) semiregular polyhedra are there? Answer: Infinitely many.

14 Prisms Definition Let n 3. An n-gonal prism is the solid obtained by connecting two congruent regular n-gons by a loop of n squares. Modified Question: How many (non-regular) semiregular polyhedra are there besides the prisms? Answer: Still infinitely many.

15 Antiprisms Definition Let n 3. An n-gonal antiprism is the solid obtained by connecting two congruent regular n-gons by a loop of 2n equilateral triangles. Re-Modified Question: How many semiregular polyhedra are there besides the regular polyhedra, prisms, and antiprisms? Answer: FINALLY that s the right question. Archimedes says: 13.

16 Archimedean Solids Definition An Archimedean solid is a semiregular polyhedron that is not regular, a prism, or an antiprism. Archimedes of Syracuse, BCE: Among his many mathematical contributions, described the 13 Archimedean solids. But this work is lost. We know of it only through: Pappus of Alexandria, c.290 c.350 CE: One of the last ancient Greek mathematicians. Describes the 13 Archimedean solids in Book V of his Collections. Johannes Kepler (Germany and Austria), CE: Rediscovered the 13 Archimedean solids; gave first surviving proof that there are only 13.

17 Cuboctahedron and Icosidodecahedron

18 Adding Some Squares Rhombicuboctahedron Rhombicosidodecahedron

19 Truncating Regular Polyhedra To truncate a polyhedron means to slice off its corners. Truncated Tetrahedron Truncated Cube

20 More Truncated Regular Polyhedra Truncated Octahedron Truncated Dodecahedron

21 Truncated Icosahedron

22 Truncation FAIL If you truncate, say, the cuboctahedron, you don t quite get regular polygons the sliced corners give non-square rectangles.

23 Fixing a Failed Truncation But if you squish the rectangles into squares, you can get regular polygons all around. Same deal for truncating the icosidodecahedron: Truncated Cuboctahedron Truncated Icosidodecahedron a.k.a. Great Rhombicuboctahedron, Great Rhombicosidodecahedron

24 And Two More Adding more triangles to the cuboctahedron (or cube) and to the icosidodecahedron (or dodecahedron) gives: Snub Cube Snub Dodecahedron Note: Unlike all the others, these two are not mirror-symmetric.

25 Why Are There Only Thirteen? Kepler s proof for Archimedean solids is similar in spirit to Theaetetus proof for Platonic solids, but of course it s longer and more complicated. I ll give a sketch based on the description of Kepler s proof in Chapter 4 of Polyhedra, by Peter Cromwell. (Cambridge U Press, 1997). Goal: Given a semiregular polyhedron S, we want to show the arrangement of polygons around each vertex agrees with one of the specific examples we already know about.

26 Some Notation To describe the arrangement of polygons at a vertex, let s write [a, b, c,...] to indicate that there s an a-gon, then a b-gon, then a c-gon, etc., as we go around the vertex. Example: The truncated cube is [8, 8, 3], and the icosidodecahedron is [3, 5, 3, 5]. Warning: the order matters, but only up to rotating. So [8, 8, 3] = [8, 3, 8] = [3, 8, 8] and [3, 5, 3, 5] = [5, 3, 5, 3] [3, 3, 5, 5].

27 Three Lemmas Lemma 1. Suppose [a, b, c] is an arrangement for a semiregular polyhedron. If a is odd, then b = c. Lemma 2. Suppose [3, 3, a, b] is an arrangement for a semiregular polyhedron. Then either a = 3 or b = 3. (Antiprism.) Lemma 3. Suppose [3, a, b, c] is an arrangement for a semiregular polyhedron. If a, c 3, then a = c.

28 Sketch of the Proof There are now a whole lot of cases to consider, depending on what sorts of polygonal faces the solid S has: 1. 2 sorts: Triangles and Squares sorts: Triangles and Pentagons sorts: Triangles and Hexagons sorts: Triangles and n-gons, with n sorts: Squares and n-gons, with n sorts: Pentagons and n-gons, with n sorts: Triangles, Squares, and n-gons, with n sorts: Triangles, m-gons, and n-gons, with n > m sorts: l-gons, m-gons, and n-gons, with n > m > l or 3 sorts, all with 6 sides each sorts of polygons. And most of these cases have multiple sub-cases.

29 Example: Case 2: Triangles and Pentagons as faces One pentagon at each vertex: [3, 3, 5]: Impossible by Lemma 1: 3 odd, 3 5. [3, 3, 3, 5]: Pentagonal Antiprism [3, 3, 3, 3, 5]: Snub Dodecahedron 5 triangles and 1 pentagon: NO; angles total > 360. Two pentagons at each vertex: [3, 5, 5] = [5, 5, 3]: Impossible by Lemma 1: 5 odd, 5 3. [3, 3, 5, 5]: Impossible by Lemma 2: 5, 5 3 [3, 5, 3, 5]: Isocidodecahedron 3 triangles and 2 pentagons: NO; angles total > pentagons at each vertex: 1 triangle(s) and 3 pentagons: NO; angles total > 360.

30 A Twist OK, so the proof is done. But what about this: This new solid has only squares and triangles for faces, and each vertex has 3 squares and 1 triangle, with the same set of angles between them. It is called the Elongated Square Gyrobicupola or Pseudorhombicuboctahedron.

31 The Pseudorhombicuboctahedron First known appearance in print: Duncan Sommerville, Rediscovered by J.C.P. Miller, (Sometimes called Miller s solid ). Did Kepler know of it? He once referred to 14 Archimedean solids, rather than 13. But it s not usually considered an Archimedean solid.

32 Two Questions 1. Why did Kepler s proof miss the PRCOH? A: Unlike regular polyhedra, a convex polyhedron with the same arrangement of regular polygons at each vertex is not completely determined by the arrangement of polygons around each vertex. 2. So why isn t the PRCOH considered semiregular? A: Because we (like the ancients) were vague about what all vertices are congruent means. Is it just the faces meeting at the vertex that look the same? Or is that the whole solid looks the same if you move one vertex to where another one was? Although terminology varies, there is general agreement that the nice class (usually called either semiregular or uniform ) should use the whole solid definition. So the PRCOH is not an Archimedean solid.

33 Time to Count Stuff Let s count how many vertices, edges, and faces these solids have: V E F V E F TH T.TH Cu T.Cu OH T.OH DH T.DH IH T.IH COH IDH RCOH RIDH T.COH T.IDH S.Cu S.DH V E F V E F n-pr 2n 3n n + 2 n-apr 2n 4n 2n + 2 Euler observes: V E + F = 2.

34 Euler s Theorem Leonhard Euler (Switzerland, Russia, Germany), CE: Among many, many, MANY other things, proved: Theorem Let S be a convex polyhedron, with V vertices, E edges, and F faces. Then V E + F = 2. Key idea of proof: If you change a polyhedron by: adding a vertex somewhere in the middle of an edge, cutting a face in two by connecting two nonadjacent vertices with a new edge, reversing either of the above two kinds of operations, or bending or stretching it, the quantity V E + F remains unchanged.

35 Regular Polyhedra Revisited Let S be a regular polyhedron with m faces, and with k regular n-gons meeting at each vertex. Then: So V = mn k, 2 = V E + F = mn k E = mn 2, F = m. mn 2 m(2n kn + 2k) + m =. 2k In particular, 2n kn + 2k > 0, i.e., 1 n + 1 k > 1 2. And it s not hard to show that the only pairs of integers (k, n) with k, n 3 that give 1 n + 1 k > 1 2 are: